返回

28 , 求一个正交变换化下列二次型成标准形:

(1)f=2x12x_{1}^{2}+3x22x_{2}^{2}+3x32x_{3}^{2}+4x2x3 ; (2)f=x12x_{1}^{2}+x32x_{3}^{2}+2x1x2-2x2x3

解:(1)

具体步骤如下:

1.写出二次型的矩阵AA

二次型的矩阵是对称矩阵,主对角线元素对应平方项系数,

非主对角线元素为交叉项系数的一半。

对于f=2x12+3x22+3x32+4x2x3f = 2x_{1}^{2} + 3x_{2}^{2} + 3x_{3}^{2} + 4x_{2}x_{3},其矩阵为:A=(200032023)A = \begin{pmatrix} 2 & 0 & 0 \\ 0 & 3 & 2 \\ 0 & 2 & 3 \end{pmatrix}

2.求矩阵AA的特征值

特征值通过求解特征方程|λIA|=0|\lambda I - A| = 0得到,其中II为单位矩阵:

|λIA|=|λ2000λ3202λ3|=(λ2)(λ26λ+5)==0|\lambda I - A| = \left| \begin{matrix} \lambda - 2 & 0 & 0 \\ 0 & \lambda - 3 & - 2 \\ 0 & - 2 & \lambda - 3 \end{matrix} \right| = (\lambda - 2)(\lambda^{2} - 6\lambda + 5) = = 0

得到特征值为λ1=1\lambda_{1} = 1λ2=2\lambda_{2} = 2λ3=5\lambda_{3} = 5

3.求对应特征值的特征向量

λ1=1\lambda_{1} = 1时:解方程组(IA)X=(100022022)X=0(I - A)X = \begin{pmatrix} - 1 & 0 & 0 \\ 0 & - 2 & - 2 \\ 0 & - 2 & - 2 \end{pmatrix}X = 0

得特征向量ξ1=(011)\xi_{1} = \begin{pmatrix} 0 \\ - 1 \\ 1 \end{pmatrix}

λ2=2\lambda_{2} = 2时:解方程组(2IA)X=(000012021)X=0(2I - A)X = \begin{pmatrix} 0 & 0 & 0 \\ 0 & - 1 & - 2 \\ 0 & - 2 & - 1 \end{pmatrix}X = 0

得特征向量ξ2=(100)\xi_{2} = \begin{pmatrix} 1 \\ 0 \\ 0 \end{pmatrix}

λ3=5\lambda_{3} = 5时:解方程组(5IA)X=(300022022)X=0(5I - A)X = \begin{pmatrix} 3 & 0 & 0 \\ 0 & 2 & - 2 \\ 0 & - 2 & 2 \end{pmatrix}X = 0

得特征向量ξ3=(011)\xi_{3} = \begin{pmatrix} 0 \\ 1 \\ 1 \end{pmatrix}

4.将特征向量单位化

不同特征值对应的特征向量已正交,只需单位化(除以自身的模长):

ξ1\xi_{1}的模长ξ1=02+(1)2+12=2\parallel \xi_{1} \parallel = \sqrt{0^{2} + ( - 1)^{2} + 1^{2}} = \sqrt{2},单位化后η1=12(011)=(01212)\eta_{1} = \frac{1}{\sqrt{2}}\begin{pmatrix} 0 \\ - 1 \\ 1 \end{pmatrix} = \begin{pmatrix} 0 \\ - \frac{1}{\sqrt{2}} \\ \frac{1}{\sqrt{2}} \end{pmatrix}

ξ2\xi_{2}的模长ξ2=12+02+02=1\parallel \xi_{2} \parallel = \sqrt{1^{2} + 0^{2} + 0^{2}} = 1,单位化后η2=(100)\eta_{2} = \begin{pmatrix} 1 \\ 0 \\ 0 \end{pmatrix}

ξ3\xi_{3}的模长ξ3=02+12+12=2\parallel \xi_{3} \parallel = \sqrt{0^{2} + 1^{2} + 1^{2}} = \sqrt{2},单位化后η3=12(011)=(01212)\eta_{3} = \frac{1}{\sqrt{2}}\begin{pmatrix} 0 \\ 1 \\ 1 \end{pmatrix} = \begin{pmatrix} 0 \\ \frac{1}{\sqrt{2}} \\ \frac{1}{\sqrt{2}} \end{pmatrix}

5.构造正交矩阵PP并写出正交变换和标准形

令正交矩阵P=[η1,η2,η3]P = \lbrack\eta_{1},\eta_{2},\eta_{3}\rbrack,即:P=(0101201212012)P = \begin{pmatrix} 0 & 1 & 0 \\ - \frac{1}{\sqrt{2}} & 0 & \frac{1}{\sqrt{2}} \\ \frac{1}{\sqrt{2}} & 0 & \frac{1}{\sqrt{2}} \end{pmatrix}

则正交变换为X=PYX = PY,即(x1x2x3)\begin{pmatrix} x_{1} \\ x_{2} \\ x_{3} \end{pmatrix}=(0101201212012)(y1y2y3)\begin{pmatrix} 0 & 1 & 0 \\ - \frac{1}{\sqrt{2}} & 0 & \frac{1}{\sqrt{2}} \\ \frac{1}{\sqrt{2}} & 0 & \frac{1}{\sqrt{2}} \end{pmatrix}\begin{pmatrix} y_{1} \\ y_{2} \\ y_{3} \end{pmatrix}),

验证正交变换 x=Pyx = Py 将二次型化为标准形

已知 f=xTAxf = x^{T}Ax,令 x=Pyx = Py,则f=(Py)TA(Py)=yT(PTAP)yf = (Py)^{T}A(Py) = y^{T}(P^{T}AP)y

PTAPP^{T}AP 应该等于对角矩阵 Λ=diag(125)\Lambda = diag(1,2,5)

检查:PTAP=(0121210001212)(200032023)(0101201212012)=[100020005]P^{T}AP = \begin{pmatrix} 0 & - \frac{1}{\sqrt{2}} & \frac{1}{\sqrt{2}} \\ 1 & 0 & 0 \\ 0 & \frac{1}{\sqrt{2}} & \frac{1}{\sqrt{2}} \end{pmatrix}\begin{pmatrix} 2 & 0 & 0 \\ 0 & 3 & 2 \\ 0 & 2 & 3 \end{pmatrix}\begin{pmatrix} 0 & 1 & 0 \\ - \frac{1}{\sqrt{2}} & 0 & \frac{1}{\sqrt{2}} \\ \frac{1}{\sqrt{2}} & 0 & \frac{1}{\sqrt{2}} \end{pmatrix} = \begin{bmatrix} 1 & 0 & 0 \\ 0 & 2 & 0 \\ 0 & 0 & 5 \end{bmatrix}

二次型化为标准形:f=y12+2y22+5y32f = y_{1}^{2} + 2y_{2}^{2} + 5y_{3}^{2}

(2)二次型f的矩阵为A=[110101011]\begin{bmatrix} 1 & 1 & 0 \\ 1 & 0 & - 1 \\ 0 & - 1 & 1 \end{bmatrix} , 它的特征多项式为

|A -λE|=|1λ100λ1011λ|\left| \begin{matrix} 1 - \lambda & 1 & 0 \\ 0 & - \lambda & - 1 \\ 0 & - 1 & 1 - \lambda \end{matrix} \right|=-(λ-2)(λ-1)(λ+1)=0

所以A的特征值为λ1=2 , λ2=1 , λ3=-1

对应特征值λ1=2 ,

解方程(A-2E)x=[110121011]\begin{bmatrix} - 1 & 1 & 0 \\ 1 & - 2 & - 1 \\ 0 & - 1 & - 1 \end{bmatrix}x=0,得单位特征向量p1=13[111]\frac{1}{\sqrt{3}}\begin{bmatrix} 1 \\ 1 \\ - 1 \end{bmatrix}

对应特征值λ2=1 ,

解方程(A-E)x=[010111010]\begin{bmatrix} 0 & 1 & 0 \\ 1 & - 1 & - 1 \\ 0 & - 1 & 0 \end{bmatrix}x=0,得单位特征向量p2=13[101]\frac{1}{\sqrt{3}}\begin{bmatrix} 1 \\ 0 \\ 1 \end{bmatrix}

对应特征值λ3=-1 ,

解方程(A+E)x=[210111012]\begin{bmatrix} 2 & 1 & 0 \\ 1 & 1 & - 1 \\ 0 & - 1 & - 2 \end{bmatrix}x=0,得单位特征向量p3=16[121]\frac{1}{\sqrt{6}}\begin{bmatrix} - 1 \\ 2 \\ 1 \end{bmatrix}

令P=(p1 , p2 , p3)=[13121613026131216]\begin{bmatrix} \frac{1}{\sqrt{3}} & \frac{1}{\sqrt{2}} & \frac{- 1}{\sqrt{6}} \\ \frac{1}{\sqrt{3}} & 0 & \frac{2}{\sqrt{6}} \\ \frac{- 1}{\sqrt{3}} & \frac{1}{\sqrt{2}} & \frac{1}{\sqrt{6}} \end{bmatrix} , 则P为正交矩阵 , 正交变换为x=Py

[x1x2x3]\begin{bmatrix} x_{1} \\ x_{2} \\ x_{3} \end{bmatrix}=[13121613026131216][y1y2y3]\begin{bmatrix} \frac{1}{\sqrt{3}} & \frac{1}{\sqrt{2}} & \frac{- 1}{\sqrt{6}} \\ \frac{1}{\sqrt{3}} & 0 & \frac{2}{\sqrt{6}} \\ \frac{- 1}{\sqrt{3}} & \frac{1}{\sqrt{2}} & \frac{1}{\sqrt{6}} \end{bmatrix}\begin{bmatrix} y_{1} \\ y_{2} \\ y_{3} \end{bmatrix} 或X: {x1=13y1+12y216y3x2=13y1+26y3x3=13y1+12y2+16y3\left\{ \begin{matrix} x_{1} = \frac{1}{\sqrt{3}}y_{1} + \frac{1}{\sqrt{2}}y_{2} - \frac{1}{\sqrt{6}}y_{3} \\ x_{2} = \frac{1}{\sqrt{3}}y_{1} + \frac{2}{\sqrt{6}}y_{3}\ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \\ x_{3} = \frac{- 1}{\sqrt{3}}y_{1} + \frac{1}{\sqrt{2}}y_{2} + \frac{1}{\sqrt{6}}y_{3} \end{matrix} \right.\

验证正交变换 x=Pyx = Py 将二次型化为标准形

已知 f=xTAxf = x^{T}Ax,令 x=Pyx = Py,则f=(Py)TA(Py)=yT(PTAP)yf = (Py)^{T}A(Py) = y^{T}(P^{T}AP)y

PTAPP^{T}AP 应该等于对角矩阵 Λ=diag(211)\Lambda = diag(2,1, - 1)

检查:PTAP=(13131312012162616)[110101011][13121613026131216]=[200010001]P^{T}AP = \begin{pmatrix} \frac{1}{\sqrt{3}} & \frac{1}{\sqrt{3}} & \frac{- 1}{\sqrt{3}} \\ \frac{1}{\sqrt{2}} & 0 & \frac{1}{\sqrt{2}} \\ \frac{- 1}{\sqrt{6}} & \frac{2}{\sqrt{6}} & \frac{1}{\sqrt{6}} \end{pmatrix}\begin{bmatrix} 1 & 1 & 0 \\ 1 & 0 & - 1 \\ 0 & - 1 & 1 \end{bmatrix}\begin{bmatrix} \frac{1}{\sqrt{3}} & \frac{1}{\sqrt{2}} & \frac{- 1}{\sqrt{6}} \\ \frac{1}{\sqrt{3}} & 0 & \frac{2}{\sqrt{6}} \\ \frac{- 1}{\sqrt{3}} & \frac{1}{\sqrt{2}} & \frac{1}{\sqrt{6}} \end{bmatrix}\ = \begin{bmatrix} 2 & 0 & 0 \\ 0 & 1 & 0 \\ 0 & 0 & - 1 \end{bmatrix}

所以f的标准形f=2y12y_{1}^{2}+y22y_{2}^{2}-y32y_{3}^{2}

29 , 求一个正交变换

把二次曲面的方程3x2+5y2+5z2+4xy-4xz-10yz=1化成标准方程

解:记二次曲面为f=1 , 则f为二次型 , 它的矩阵为A=[322255255]\begin{bmatrix} 3 & 2 & - 2 \\ 2 & 5 & - 5 \\ - 2 & - 5 & 5 \end{bmatrix}

它的特征多项式为

|A -λE|=|3λ2225λ5255λ|\left| \begin{matrix} 3 - \lambda & 2 & - 2 \\ 2 & 5 - \lambda & - 5 \\ - 2 & - 5 & 5 - \lambda \end{matrix} \right|=(-λ)(λ-2)(λ-11)=0

所以A的特征值为λ1=0 , λ2=2 , λ3=11

对应特征值λ1=0 ,

解方程Ax=[322255255]\begin{bmatrix} 3 & 2 & - 2 \\ 2 & 5 & - 5 \\ - 2 & - 5 & 5 \end{bmatrix}x=0,得单位特征向量p1=12[011]\frac{1}{\sqrt{2}}\begin{bmatrix} 0 \\ 1 \\ 1 \end{bmatrix}

对应特征值λ2=2 ,

解方程(A-2E)x=[122235253]\begin{bmatrix} 1 & 2 & - 2 \\ 2 & 3 & - 5 \\ - 2 & - 5 & 3 \end{bmatrix}x=0,得单位特征向量p2=132[411]\frac{1}{3\sqrt{2}}\begin{bmatrix} 4 \\ - 1 \\ 1 \end{bmatrix}

对应特征值λ3=11 ,

解方程(A-11E)x=[822265256]\begin{bmatrix} - 8 & 2 & - 2 \\ 2 & - 6 & - 5 \\ - 2 & - 5 & - 6 \end{bmatrix}x=0,得单位特征向量p3=13[122]\frac{1}{3}\begin{bmatrix} 1 \\ 2 \\ - 2 \end{bmatrix}

令P=(p1 , p2 , p3)=[04321312132231213223]\begin{bmatrix} 0 & \frac{4}{3\sqrt{2}} & \frac{1}{3} \\ \frac{1}{\sqrt{2}} & \frac{- 1}{3\sqrt{2}} & \frac{2}{3} \\ \frac{1}{\sqrt{2}} & \frac{1}{3\sqrt{2}} & \frac{- 2}{3} \end{bmatrix} , 则P为正交阵 , 并且正交变换是

[xyz]\begin{bmatrix} x \\ y \\ z \end{bmatrix}=[04321312132231213223][uvw]\begin{bmatrix} 0 & \frac{4}{3\sqrt{2}} & \frac{1}{3} \\ \frac{1}{\sqrt{2}} & \frac{- 1}{3\sqrt{2}} & \frac{2}{3} \\ \frac{1}{\sqrt{2}} & \frac{1}{3\sqrt{2}} & \frac{- 2}{3} \end{bmatrix}\begin{bmatrix} u \\ v \\ w \end{bmatrix}{x=432v+13wy=12u132v+23wz=12u+132v23w\left\{ \begin{matrix} x = \frac{4}{3\sqrt{2}}v + \frac{1}{3}w\ \ \ \ \ \ \ \ \ \ \ \ \ \\ y = \frac{1}{\sqrt{2}}u - \frac{1}{3\sqrt{2}}v + \frac{2}{3}w \\ z = \frac{1}{\sqrt{2}}u + \frac{1}{3\sqrt{2}}v - \frac{2}{3}w \end{matrix} \right.\

当是正交变换时,标准形系数就是特征值

所以二次曲面的标准方程为2v2+11w2=1(它是椭圆柱面)

验证:

设a=132\frac{1}{3\sqrt{2}} , b=13\frac{1}{3} , c=12\frac{1}{\sqrt{2}} , d=23\frac{2}{3}

{x=432v+13wy=12u132v+23wz=12u+132v23w\left\{ \begin{matrix} x = \frac{4}{3\sqrt{2}}v + \frac{1}{3}w\ \ \ \ \ \ \ \ \ \ \ \ \ \\ y = \frac{1}{\sqrt{2}}u - \frac{1}{3\sqrt{2}}v + \frac{2}{3}w \\ z = \frac{1}{\sqrt{2}}u + \frac{1}{3\sqrt{2}}v - \frac{2}{3}w \end{matrix} \right.\ 变为{x=4av+bwy=cuav+dwz=cu+avdw\left\{ \begin{matrix} x = 4av + bw\ \ \ \ \ \ \ \ \ \\ y = cu - av + dw \\ z = cu + av - dw \end{matrix} \right.\

{x=4av+bwy=cuav+dwz=cu+avdw\left\{ \begin{matrix} x = 4av + bw\ \ \ \ \ \ \ \ \\ y = cu - av + dw \\ z = cu + av - dw \end{matrix} \right.\ 代入3x2+5y2+5z2+4xy-4xz-10yz=1

化简得36a²v² + (3b² + 20d² + 8bd)w² + (16ab - 8ad)vw = 1

把a=132\frac{1}{3\sqrt{2}} , b=13\frac{1}{3} , c=12\frac{1}{\sqrt{2}} , d=23\frac{2}{3}

代入36a²v² + (3b² + 20d² + 8bd)w² + (16ab - 8ad)vw = 1

并化简得2v2+11w2=1

验证正交变换 x=Pyx = Py 将二次型化为标准形

已知 f=xTAxf = x^{T}Ax,令 x=Pyx = Py,则f=(Py)TA(Py)=yT(PTAP)yf = (Py)^{T}A(Py) = y^{T}(P^{T}AP)y

PTAPP^{T}AP 应该等于对角矩阵 Λ=diag(0211)\Lambda = diag(0,2,11)

检查:PTAP=[01212432132132132323][322255255][04321312132231213223]=[0000200011]P^{T}AP = \begin{bmatrix} 0 & \frac{1}{\sqrt{2}} & \frac{1}{\sqrt{2}} \\ \frac{4}{3\sqrt{2}} & \frac{- 1}{3\sqrt{2}} & \frac{1}{3\sqrt{2}} \\ \frac{1}{3} & \frac{2}{3} & \frac{- 2}{3} \end{bmatrix}\ \begin{bmatrix} 3 & 2 & - 2 \\ 2 & 5 & - 5 \\ - 2 & - 5 & 5 \end{bmatrix}\begin{bmatrix} 0 & \frac{4}{3\sqrt{2}} & \frac{1}{3} \\ \frac{1}{\sqrt{2}} & \frac{- 1}{3\sqrt{2}} & \frac{2}{3} \\ \frac{1}{\sqrt{2}} & \frac{1}{3\sqrt{2}} & \frac{- 2}{3} \end{bmatrix}\ \ = \begin{bmatrix} 0 & 0 & 0 \\ 0 & 2 & 0 \\ 0 & 0 & 11 \end{bmatrix}

30 , 证明二次型f=xTAx在x\left\| x \right\|=1时的最大值为矩阵A的最大特征值

证:

证明过程 前提说明:

二次型的矩阵AAnn阶实对称矩阵,因此其特征值均为实数,

且可通过正交变换实现对角化,这是证明的基础。

步骤1:对实对称矩阵AA做正交对角化

根据实对称矩阵的性质,存在nn阶正交矩阵PP

使得:PTAP=Λ=diag(λ1,λ2,,λn)P^{T}AP = \Lambda = diag(\lambda_{1},\lambda_{2},\ldots,\lambda_{n})

其中λ1,λ2,,λn\lambda_{1},\lambda_{2},\ldots,\lambda_{n}是矩阵AA的全部特征值,且均为实数。

λmax=max{λ1,λ2,,λn}\lambda_{\max} = max\{\lambda_{1},\lambda_{2},\ldots,\lambda_{n}\},即AA的最大特征值。

步骤2:利用正交变换化简二次型,且保持向量模长不变

对向量xx做正交变换x=Pyx = Py,其中y=(y1,y2,,yn)Ty = (y_{1},y_{2},\ldots,y_{n})^{T}

正交变换保范数:因PP是正交矩阵,满足PTP=EP^{T}P = E

因此x=xTx=(Py)T(Py)=yTPTPy=yTy=y\parallel x \parallel = \sqrt{x^{T}x} = \sqrt{(Py)^{T}(Py)} = \sqrt{y^{T}P^{T}Py} = \sqrt{y^{T}y} = \parallel y \parallel

故约束条件x=1\parallel x \parallel = 1等价于y=1\parallel y \parallel = 1,即y12+y22++yn2=1y_{1}^{2} + y_{2}^{2} + \ldots + y_{n}^{2} = 1

二次型化简:将x=Pyx = Py代入二次型,

f=xTAx=(Py)TA(Py)=yT(PTAP)y=yTΛyf = x^{T}Ax = (Py)^{T}A(Py) = y^{T}(P^{T}AP)y = y^{T}\Lambda y

展开后为:f=λ1y12+λ2y22++λnyn2f = \lambda_{1}y_{1}^{2} + \lambda_{2}y_{2}^{2} + \ldots + \lambda_{n}y_{n}^{2}

步骤3:证明二次型的上界为λmax\lambda_{\max}

对所有i=1,2,,ni = 1,2,\ldots,n,有λiλmax\lambda_{i} \leq \lambda_{\max}

因此对二次型做放缩:f=i=1nλiyi2i=1nλmaxyi2=λmaxi=1nyi2f = \sum_{i = 1}^{n}\lambda_{i}y_{i}^{2} \leq \sum_{i = 1}^{n}\lambda_{\max}y_{i}^{2} = \lambda_{\max} \cdot \sum_{i = 1}^{n}y_{i}^{2}

结合约束条件y=1\parallel y \parallel = 1,即i=1nyi2=1\sum_{i = 1}^{n}y_{i}^{2} = 1,因此:fλmaxf \leq \lambda_{\max}

步骤4:证明上界λmax\lambda_{\max}可以取到(可达性)

ξ\xiAA对应于特征值λmax\lambda_{\max}的单位特征向量,

即满足:Aξ=λmaxξ,ξ=1A\xi = \lambda_{\max}\xi,\quad \parallel \xi \parallel = 1

x=ξx = \xi代入二次型,

得:f=ξTAξ=ξT(λmaxξ)=λmaxξTξ=λmaxξ2=λmaxf = \xi^{T}A\xi = \xi^{T}(\lambda_{\max}\xi) = \lambda_{\max} \cdot \xi^{T}\xi = \lambda_{\max} \cdot \parallel \xi \parallel^{2} = \lambda_{\max}

这说明存在满足x=1\parallel x \parallel = 1的向量x=ξx = \xi,使得二次型ff取到λmax\lambda_{\max}

结论 综上,二次型f=xTAxf = x^{T}Axx=1\parallel x \parallel = 1时的最大值,

恰好为矩阵AA的最大特征值λmax\lambda_{\max}

同理可证,该二次型在x=1\parallel x \parallel = 1时的最小值,为矩阵AA的最小特征值。

举例:

1. 对于对称矩阵A=(3113)A = \begin{pmatrix} 3 & 1 \\ 1 & 3 \end{pmatrix}

特征值 λ1=4\lambda_{1} = 4λ2=2\lambda_{2} = 2,最大特征值 44

2. 取一些模长为 1 的向量

1. 取 x=(10)x = \begin{pmatrix} 1 \\ 0 \end{pmatrix}x=1\parallel x \parallel = 1

f=xTAx=(1,0)(3113)(10)=(1,0)(3,1)=3x^{T}Ax = (1,0)\begin{pmatrix} 3 & 1 \\ 1 & 3 \end{pmatrix}\begin{pmatrix} 1 \\ 0 \end{pmatrix} = (1,0) \cdot (3,1) = 3

2. 取 x=(01)x = \begin{pmatrix} 0 \\ 1 \end{pmatrix}x=1\parallel x \parallel = 1

f=xTAx=(0,1)(3113)(01)=(0,1)(1,3)=3x^{T}Ax = (0,1)\begin{pmatrix} 3 & 1 \\ 1 & 3 \end{pmatrix}\begin{pmatrix} 0 \\ 1 \end{pmatrix} = (0,1) \cdot (1,3) = 3

3. 取 x=12(11)x = \frac{1}{\sqrt{2}}\begin{pmatrix} 1 \\ 1 \end{pmatrix}x=1\parallel x \parallel = 1

f=xTAx=12(1,1)(3113)12(11)=12(1,1)(4,4)=12×8=4x^{T}Ax = \frac{1}{\sqrt{2}}(1,1)\begin{pmatrix} 3 & 1 \\ 1 & 3 \end{pmatrix}\frac{1}{\sqrt{2}}\begin{pmatrix} 1 \\ 1 \end{pmatrix} = \frac{1}{2}(1,1) \cdot (4,4) = \frac{1}{2} \times 8 = 4

4. 取 x=12(11)x = \frac{1}{\sqrt{2}}\begin{pmatrix} 1 \\ - 1 \end{pmatrix}x=1\parallel x \parallel = 1

f=xTAx=12(1,1)(3113)(11)=12(1,1)(2,2)=12×4=2x^{T}Ax = \frac{1}{2}(1, - 1)\begin{pmatrix} 3 & 1 \\ 1 & 3 \end{pmatrix}\begin{pmatrix} 1 \\ - 1 \end{pmatrix} = \frac{1}{2}(1, - 1) \cdot (2, - 2) = \frac{1}{2} \times 4 = 2

5. 取 x=(0.60.8)x = \begin{pmatrix} 0.6 \\ 0.8 \end{pmatrix}x=0.36+0.64=1\parallel x \parallel = \sqrt{0.36 + 0.64} = 1

f=xTAx=(0.6,0.8)(3113)(0.60.8)=3.96x^{T}Ax = (0.6,0.8)\begin{pmatrix} 3 & 1 \\ 1 & 3 \end{pmatrix}\begin{pmatrix} 0.6 \\ 0.8 \end{pmatrix} = 3.96

3. 比较这些值

尝试的向量 xx 对应 f(x)=xTAxf(x) = x^{T}Ax 的值分别为: 3,3,4,2,3.96。

最大值为 44,正好等于 AA 的最大特征值,并且是在特征向量12(11)\frac{1}{\sqrt{2}}\begin{pmatrix} 1 \\ 1 \end{pmatrix} 处取到的。

其他任意模长 1 的向量算出的值都不会超过 4。

结论:用具体的模长为 1 的向量计算二次型,验证了最大值等于最大特征值 4。